基于单神经元神经网络的无刷直流电机控制系统仿真  

Simulation of Brushless Direct Current Motor Control System Based on Single Neuron Neural Network

作　　者：尹洪桥 易文俊[1] 贾芳 李璀璀 王康健 YIN Hong-qiao;YI Wen-jun;JIA Fang;LI Cui-cui;WANG Kang-jian(National Key Laboratory of Transient Physics,Nanjing University of Science and Technology,Nanjing 210094,China;304 Factory of China North Industries Group Corporation,Changzhi 046012,China)

机构地区：[1]南京理工大学瞬态物理国家重点实验室,南京210094 [2]中国兵器工业集团第304厂,长治046012

出　　处：《科学技术与工程》2021年第7期2747-2753,共7页Science Technology and Engineering

基　　金：国家自然科学基金(11472136);江苏省研究生培养创新工程(KYCX19_0338)。

摘　　要：无刷直流电机是一种多变量、非线性、参数时变以及强耦合的复杂系统,利用传统比例积分微分(proportional integral differential, PID)算法控制无刷直流电机存在参数调整困难、自适应能力差、控制精度低以及抗干扰能力弱等问题。为实现无刷直流电机的高精度控制,在转速环中引入了基于单神经元神经网络PID控制算法,研究了无刷直流电机的数学模型及运行特性,提出了单神经元神经网络PID算法,最后比较分析了在电机双闭环控制系统中转速环采用不同控制算法下的转速阶跃函数响应,以及三相电流、反电动势和电磁转矩的运行状态。结果表明:单神经元神经网络PID算法控制下的无刷直流电机其转速的阶跃函数响应具有更快的上升时间、更小的超调量以及更加稳定的运行状态。The brushless direct current motor(BLDCM)is a complex system with multi-variable,nonlinear,time-varying parameters and strong coupling.The traditional double closed loop proportional integral differential(PID)algorithm has certain problem to drive BLDCM such as bad parameter tuning,poor adaptability,low control accuracy and weak anti-interference ability.In order to achieve high precision control for BLDCM,a single neuron neural network PID algorithm was proposed for motor speed loop control.The mathematical model of BLDCM was studied by the single neuron neural network PID algorithm.Then,the operation characteristics were analyzed based on this system.Finally,the speed step function response,the operation state of three-phase current,back electromotive force(EMF)and electromagnetic torque were compared and analyzed,respectively.The results show that the speed step function response controlled by the single neuron neural network PID algorithm has a faster rise time,a smaller overshoot and a more stable operation state.

关 键 词：无刷直流电机 转速环 传统PID控制 单神经元神经网络PID算法 

分 类 号：TP273.2[自动化与计算机技术—检测技术与自动化装置]